One reason price might follow difficulty is that mining should not be too profitable (because nothing should be too profitable, the world doesn't leave free money lying around). Therefore the price of Bitcoins can't rise too much above the cost of mining (counting equipment depreciation among the costs of course). The cost of mining is proportional to the difficulty (approximately). Therefore we might expect to see price proportional to difficulty.

We do see a nearly proportional relationship in the 1st graph, but that data set was incomplete. I'd like to see that last graph redone with a linear difficulty scale so we could see how the proportionality holds up with more data.